Reverie: Low Pass Filter-Based Switch Buffer Sharing for Datacenters with RDMA and TCP Traffic
Vamsi Addanki, Wei Bai, Stefan Schmid, Maria Apostolaki
Abstract
The switch buffers in datacenters today are dynamically shared by traffic classes with different loss tolerance and reaction to congestion signals. In particular, while legacy applications use loss-tolerant transport, e.g., DCTCP, newer applications require lossless datacenter transport, e.g., RDMA over Converged Ethernet. Unfortunately, as we analytically show in this paper, the buffer-sharing practices of today's datacenters pose a fundamental limitation to effectively isolate RDMA and TCP while also maximizing burst absorption. We identify two root causes: (i) the buffer-sharing for RDMA and TCP relies on two independent and often conflicting views of the buffer, namely ingress and egress; and (ii) the buffer-sharing scheme micromanages the buffer and overreacts to the changes in its occupancy during transient congestion.
In this paper, we present REVERIE, a buffer-sharing scheme, which, unlike prior works, is suitable for both lossless and losstolerant traffic, providing isolation and better burst absorption than state-of-the-art buffer-sharing schemes. At the core of REVERIE lies a unified (consolidated ingress and egress) admission control that jointly optimizes the buffers for both RDMA and TCP. REVERIE allocates buffer based on a lowpass filter that naturally absorbs bursty queue lengths during transient congestion within the buffer limits. Our evaluation shows that REVERIE can improve the performance of RDMA as well as TCP in terms of flow completion times by up to 33%.
Ask about this paper
Your agent reads all of it.
Lune indexed this paper to the last equation, along with the top-tier papers that cite it. Ask a question and the answer quotes them.
Your agent calls
Luneget_paper_fulltext
Free to start. No credit card required.
Terminal
Install the CLIlune papers fulltext 51ce3452-b216-4bb8-8713-a75fc04a5b27Cited by top-tier papers9
- Credence: Augmenting Datacenter Switch Buffer Sharing with ML PredictionsVamsi Addanki, Maciej Pacut, Stefan SchmidNSDI 2024 · 22 citations
- Zoom2Net: Constrained Network Telemetry ImputationFengchen Gong, Divya Raghunathan, Aarti Gupta, Maria ApostolakiSIGCOMM 2024 · 11 citations
- Enabling Virtual Priority in Data Center Congestion ControlZhaochen Zhang, Feiyang Xue, Keqiang He, Zhimeng Yin et al.EuroSys 2025 · 6 citations
- Occamy: A Preemptive Buffer Management for On-chip Shared-memory SwitchesDanfeng Shan, Yunguang Li, Jinchao Ma, Zhenxing Zhang et al.EuroSys 2025 · 5 citations
- Efficient and Flexible Datapaths for Fine-Grained Rack-Scale Interconnects with Elastic QPChenxingyu Zhao, Yibo Wu, Hongtao Zhang, Jaehong Min et al.SIGCOMM 2026 · 1 citation
Builds on12
- When Cloud Storage Meets RDMAYixiao Gao, Qiang Li, Lingbo Tang, Yongqing Xi et al.NSDI 2021 · 228 citations
- Empowering Azure Storage with RDMAWei Bai, Shanim Sainul Abdeen, Ankit Agrawal, Krishan Kumar Attre et al.NSDI 2023 · 117 citations
- PowerTCP: Pushing the Performance Limits of Datacenter NetworksVamsi Addanki, Oliver Michel, Stefan SchmidNSDI 2022 · 116 citations
- Collie: Finding Performance Anomalies in RDMA SubsystemsXinhao Kong, Yibo Zhu, Huaping Zhou, Zhuo Jiang et al.NSDI 2022 · 86 citations
- Annulus: A Dual Congestion Control Loop for Datacenter and WAN Traffic AggregatesAhmed Saeed, Varun Gupta, Prateesh Goyal, Milad Sharif et al.SIGCOMM 2020 · 72 citations
Related papers
- FLB: Fine-grained Load Balancing for Lossless Datacenter NetworksJinbin Hu, Wenxue Li, Xiangzhou Liu, Junfeng Wang et al.USENIX ATC 2025 · 10 citations
- ABM: active buffer management in datacentersVamsi Addanki, Maria Apostolaki, Manya Ghobadi, Stefan Schmid et al.SIGCOMM 2022 · 59 citations
- Network Load Balancing with In-network Reordering Support for RDMACha Hwan Song, Xin Zhe Khooi, Raj Joshi, Inho Choi et al.SIGCOMM 2023 · 110 citations
- PACC: Proactive and Accurate Congestion Feedback for RDMA Congestion ControlXiaolong Zhong, Jiao Zhang, Yali Zhang, Zixuan Guan et al.INFOCOM 2022 · 36 citations
- InfiniFlow: Decoupling Virtual Channel Scalability from Buffer Requirements in Lossless Datacenter NetworksZerui Tian, Sen Liu, Minkun Xue, Hao Shangguan et al.SIGCOMM 2026
